Robust Lean Blowout Sensing for Turbine Engine Combustors

Abstract

Lean blowout (LBO) proximity detection is demonstrated in a Jet-A fueled, single-element Lean Direct Injection (LDI) combustor operating at conditions intended to simulate supersonic cruise (2–4 atm, ∼700 K inlet air). Detection is based on identifying LBO precursors in the optical radiation produced by the combustion process. The precursors result from short duration partial flame extinction and re-ignition events. Thresholding of the low-pass filtered optical signal provides robust precursor identification in the presence of the moderate levels of combustion dynamics exhibited by the combustor. Acoustic-based precursor sensing was also attempted, but the precursors could not be reliably identified in the presence of the combustion noise and dynamics; likely causes are discussed. The average event occurrence rate in the optical signal provides a reliable measure of the combustor’s proximity to LBO, monotonically increasing as the LBO limit is approached. The precursor events occur intermittently, and the occurrence rate statistics are reasonably modeled with a normal probability distribution. A probabilistic approach is developed to estimate the fastest fuel reduction transients that could be controlled based on the need to detect some events before the blowout conditions are reached. The results from the LDI combustor are also compared to tests in a premixed, swirl- and dump-stabilized gas-fueled combustor with a center body. High-speed flame images of the precursor events indicate the existence of different static stability modes in the premixed combustor, which are absent in the LDI combustor. The absence of mode-switching in the LDI combustor is suggested to be the reason why it produces much shorter but more frequent precursor events compared to the premixed combustor.